Choose a format. This will depend on the space you have to use, your budget and your desired ergonomics.
Buy at least two voltage-controlled oscillators in your chosen format. This will be the beginning of the sound chain. An oscillator is the basic sound source in a modular synthesizer.
Buy at least one voltage-controlled filter. A filter is responsible for shaping the sound. Start out with a basic low-pass filter or multimode filter. A multimode filter will give you the most flexibility.
Buy at least one audio mixer. It combines audio signals to feed into the filter.
Buy at least one envelope generator. It will allow you to create sound events that change over time.
Buy at least one voltage-controlled amplifier. It will allow you to control the level of one signal with a second signal. It usually is used in the output stage.
Buy sound processing modules, such as a ring modulator or a reverb generator, which will allow you to create more exotic sounds.
Buy utility modules, such as a slew limiter, envelope follower, sample and hold module and preamplifier. These will allow you to create complex patches.
Buy a power supply and distribution hardware to power your system. Make sure you select a power supply that can fulfill the power requirements of all your modules, but leave some unused power for future expansion.
Buy an enclosure large enough to fit all your modules with some room for expansion in the future. The enclosure will bring the modules together into a handsome system.
Buy cables to patch between modules.
Install the power hardware in the enclosure following the instructions provided with the equipment.
Install the modules in the enclosure and connect the power distribution.